In this study, we conducted a multi-omics analysis in peripheral blood cells from 480 ALS cases and 207 health control of PUTH-ALS, a Han Chinese ancestry cohort. We effectively controlled for confounding by batch effects, gender, age, geographical region along with estimated WBC and found that most of DMPs showed hypermethylation. Unusual DNAm patterns have been suggested to contribute to the initiation or progression of neurodegenerative diseases, such as ALS55. Morahan et al using Affymetrix GeneChip Human Tiling 2.0R Arrays on brain DNA from 10 sALS patients and 10 neurologically normal controls, reported 38 abnormal methylation sites56. Likewise, a methylome-wide association study (MWAS) performed in postmortem spinal cord tissue of 12 ALS subjects and 11 age and gender-matched neurologically normal controls identified 4,261 significant DMPs, mapping to 3,574 genes18. These studies also revealed significant hypermethylation of genes involved in calcium dynamics, oxidative stress, and synapses, although most of the DNA methylation changes were found in non-promoter regions (intronic and cryptic), aligning with our findings18,56. Despite the use of peripheral blood cells, in our gene ontology analysis, many diseases associated DMPs were linked to genes involved in neuronal functions such as axon guidance, dendrite development, and neuron projection guidance, further supporting the possible involvement of aberrant DNA methylation of neurodevelopmental genes in the pathogenesis of ALS.
We captured 5 candidate biomarkers based on methylation, such as ANKLE2, SSH2, CDC42BPB, ELAVL3, and CLEC14A, which exhibited significantly high methylation in ALS patients and were verified to be associated with statistically significant differential gene expression in an independent data source (NYGC-ALS postmortem brain tissue data expression profiles). Consistent with our findings, the transcriptomic profiles of motor neurones of the nervous system of sALS identified ELAVL3 as one of the genes most downregulated genes57. In mice models, Elavl3−/− mice were born healthy and viable with a normal life span, displayed slowly progressive motor deficits leading to severe cerebellar ataxia. This ataxia manifested as an abnormal step cycle, tremor and over time, impaired postural reflexes35. At the cellular level, axons of Elavl3−/− Purkinje cells were swollen (spheroid formation), and synaptic formation was disrupted at cerebellar nuclei35. A human tissue-based experiment also reported that uniform down-regulation of ELAVL3 in extensive models of ALS iPSC-MNs at both the RNA and the protein levels. Importantly, this decrease was present beginning in early development, even preceding TDP-43 abnormalities, suggesting ELAVL3 as an early modulator of downstream with potential as a new biomarker for ALS or a focus for therapeutic-targeted research48. In summary, these genes, well established in their role in the beginning or progression of neurodegenerative diseases, including ALS32–36,48, emerge as promising candidates for further exploration as potential biomarkers in ALS. We also conducted a difference analysis of the level of methylation in common pathogenic genes for ALS, including SOD1, TARDBP, FUS, and C9orf72, but the methylation levels of these genes did not show statistically significant differences.
Additionally, following the adjustment for known confounders, we identified 3 DMPs corresponding to RRP1B, IL17RD, and FBXO11, which were associated with survival time in 203 PUTH-ALS cases. Rrp1b has been found to be associated with survival in motor neuron disease (MND) in mice, and predicted to be deleterious and/or damaging to protein function, thus supporting our findings58. These results also suggest that these DMPs could serve as potential indicators of the underlying disease processes, offering opportunities for therapeutic interventions.
Most importantly, we identified a signature of 27 loci that can assess the risk of ALS based on abnormal methylation patterns by using machine learning algorithms. A DNA methylation association analysis based on Illumina 450K data from whole blood for an Australian ALS case–control cohort (782 cases and 613 controls) identified a case–control status classifier including the 25 most-associated probes using mixed linear models, and achieved a maximum AUC of 0.6955. In contrast, we used machine learning methods that were more suitable for processing high-dimensional data to identify a signature, which greatly improved the discrimination ability of the case-control status classifier (AUC = 0.861). Also, a considerable number (10/27) of the 27 sites encompassed in this signature overlapped substantially with the top 20 DMPs we identified after GC correction. Interestingly, the AUC was much higher for DNAm classification compared to the prediction of the SNP polygenic risk score. Considering the interaction between methylation and genetic profile scores, our results demonstrated that the interaction model explained a slightly larger proportion of variance compared to additive and signal models (0.13% for genetic profile scores, 16.99% for methylation profile scores, 17.06% for additive model, 17.08% for interaction model). However, the genetic risk score was calculated using GWAS statistics summary from the European population. In the future, the collection and integration of ALS case data in China should be accelerated to further improve prediction accuracy. In conclusion, our findings both complement and extend epigenomics research in the field of ALS in China, holding substantial promise for advancing the early clinical diagnosis of ALS.
To our knowledge, this is the largest EWAS of ALS reported to date in the Han Chinese population. Our major strength is the sample size that enabled the detection of a replicable epigenome-wide significant locus, which suggests that in blood, DNA methylation signatures associated with ALS may be subtle and will require large samples to be detected. However, this study has several limitations. First, the study used peripheral blood tissue for DNA methylation profiling, some DNA methylation studies have examined postmortem brain tissues, which are both costly and difficult to obtain. Although peripheral blood is not considered to be the most relevant tissue for the pathophysiology of ALS, our observation of significantly enriched GO terms for the DMP-related genes revealed that the development of certain brain regions could be reflected in the DNA methylome of blood samples. This observation indicated that specific epigenetic markers in brain tissue can be mirrored by the corresponding sites in peripheral blood samples. Therefore, peripheral blood could serve as a valuable surrogate for brain tissue to meet the needs of large-scale or longitudinal studies. Second, since our data comes from the Peking University Third Hospital, more samples are from the north China, and we need to conduct a larger-scale North-South balanced sample study. Third, although we adjusted for potential confounders, the possibility of residual confounding cannot be excluded. Epigenetics are vulnerable to environmental influences, such as the patient's smoking, drug use and so on. It may affect the relationship between DNA methylation and outcome variables, and more detailed environmental variables also need further consideration.